claude-mx-test SKILL.md skillB
claude-mx-test from bluechonk/modelmux is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add bluechonk/modelmux/claude-mx-test@git:20261107.0bf381dPins git:20261107.0bf381d in mdr.lock. Pin the fingerprint instead.
- Audit grade
- B, 16 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- bluechonk/modelmux
- Path
.agents/skills/claude-mx-test/SKILL.md- Versions recorded
- 2
- Latest version
- git:20261107.0bf381d, committed Nov 7, 2026
- First version recorded
- Nov 7, 2026
- Size
- 4,723 bytes
- Repository stars
- 11
- License
- MIT
用 Claude Code 对 modelmux（mx）网关做端到端测试（Anthropic Messages 协议，/v1/messages + count_tokens）。适用于：用 claude 测 mx、claude code 接入 modelmux、验证 tool_use ↔ tool_calls 双向映射与 tool_result 回灌、协议端到端验收、换协议复测模型、排查 claude 到 mx 链路报错。模型家族裸名由网关解析（claude-* 自动落池主模型），落点渠道由网关按额度挑；测试前后要做账单核对。.
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# Claude Code → modelmux 端到端测试（Anthropic Messages） 用 Claude Code CLI 直连 modelmux 网关（`/v1/messages` + `/v1/messages/count_tokens`）， 对池内模型做端到端 Agent 测试。协议转换的事实与已知边界见 [docs/CLIENT-INTEGRATION.md](../../../docs/CLIENT-INTEGRATION.md)； 本文件只保留**流程、任务矩阵与判据**。 ## 测试成本红线 端到端 Agent 测试一轮几万 token，高倍率模型等于直接烧钱。对外模型只有家族表里那几个 （`deepseek-v4.1-flash` / `glm-5.3-flash`），落点渠道由网关按额度挑；每次测试前后用 `mx bill <cid>` 核对账单（预期差值 > 0 且量级合理）。不确定就先 `mx model list` 把 选择和成本说清楚，不要直接开跑。 ## 前置条件 | 条件 | 验证 | | --- | --- | | 网关在跑 | `curl -s http://127.0.0.1:4412/health` 含 `"ok":true` 与 `"service":"modelmux"` | | 接入已就绪 | 仪表盘「接入 claude code」点过（写 `~/.claude/settings.json` 的 env），或用下面的隔离方式 | | 至少一个渠道已登录 | `mx status --json` 里该渠道 `credentials.present` 为 `true` | **隔离方式**（不动真实配置，验证链路时首选）： ```bash CLAUDE_CONFIG_DIR=<临时目录> \ ANTHROPIC_BASE_URL=http://127.0.0.1:4412 \ ANTHROPIC_AUTH_TOKEN=modelmux-local \ ANTHROPIC_MODEL=deepseek-v4.1-flash \ claude -p "<任务>" ``` ## 测试任务矩阵（按深度递进，全部实测过） 每发之间查一次 `mx log --lines 3`：`protocol` 必须是 `anthropic`、`status 200`、 `outcome success`。工具任务在仓库根跑，落盘产物进 `.tmp/`（用完即删，不进 git）。 1. **冒烟**：`claude -p "只回答一个词：pong"` → 返回 `pong`。 开头的 `[claude-code:unrecognized_model]` 是 Claude Code 自己的提示日志 …
Read the whole file at its exact version.
How to install
mdr add bluechonk/modelmux/claude-mx-test@sha256:317a07ce6e04039dThe line at the top of this page pins the label git:20261107.0bf381d, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_nqblisxcznkz6ozg)
2 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 2 versions of this file, all committed on Nov 7, 2026.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| git:20261107.0bf381d latest | 2026-11-07 | 0bf381d | 4,723 B | B | view · diff |
| git:20261107.bd0e3d9 | 2026-11-07 | bd0e3d9 | 4,721 B | B | view |
Audit of the latest version
- fail: No link to a raw IP address (matched: http://127.0.0.1:4412)
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (4723 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No script tag
Starting a new one? Use a template for SKILL.md on mdpantry, which passes every one of these checks and says what to change in it.
Source
bluechonk/modelmux · 11 stars · license MIT · pushed 2026-11-07 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_nqblisxcznkz6ozg GET https://markdownregistry.com/api/v1/resolve?ref=bluechonk/modelmux/claude-mx-test GET https://markdownregistry.com/api/v1/blob/317a07ce6e04039dfc1161999f53a4e322793c65ec2c6ea60892c7343d381917
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.